Auto-Converting V4 Games to V5

[David Whyld]

In testing a work in progress of mine recently, I noticed that the game plays quite differently in V5 than it does in V4 (which it was written in). It throws up error messages quite frequently which I'm not entirely sure how to fix, and default messages I had overridden in the ALR are back to their original state. Anyone playing the game with the V5 Runner is going to run into all manner of problems with it, and I'm guessing the same will apply to anyone trying to play it via the WebRunner*.

Might it be an idea to warn anyone playing a V4 game with the V5 Runner that they're better off running it with the V4 Runner instead? And disable WebRunner play for V4 games by default? It's not doing ADRIFT any favours if the games are buggy because the player is using the wrong version of the software to run them.

* Which, incidentally, is throwing up the following error message when I try to run the latest V5 game on the main site: "The file is a later version than the software. It is advisable that you upgrade to ensure it runs properly." That's a pretty offputting error message.

[Campbell]

I've fixed the WebRunner error.

Regarding differences between v4 and v5, I would rather get these corrected, so if you are able to send me the game in question and let me know where they differ, I will see if I can get that fixed.

[ralphmerridew]

That may be easier said than done; my experience is that trying to translate one high-level language to another, converting high-level concepts to high-level concepts, generally won't give full compatibility; you have to emulate the high-level concepts with low-level ones. The ALR is complex enough in its interactions that you'll have to duplicate the exact sequence of variable evaluation and text replacement.

Here's an edge case I constructed a while back: http://pastebin.com/mBuzgH8u
Go to either room, and type "reset", then "advance" a few times.

Under ADRIFT 4, in the west room, the counter starts from 0; in the east room, the counter starts from 1, even though the only difference between the rooms is Task2, which does nothing.

Campbell will probably understand why this quirk happens, but I doubt any other drifter will. I think that the only method that will work will be:
1) Include a separate core in the 5.0 Runner for running 4.0 games.
2) View the 4.0 -> 5.0 conversion as being the first step, and that it requires completely retesting the game.

[David Whyld]

Campbell, I don't mind sending the game to you so you can take a look at it, but I'm not sure that's really going to solve anything in the long term. Unless you can 100% guarantee that every V4 game will run flawlessly with V5 – something which I doubt is even possible given the sheer number of things that would need to be covered – there are always going to be problems running V4 games with V5, and as such I think it makes sense to at least warn people playing the games that they're using the wrong version of the software.

I remember playing V3.9 games with V4 and running into problems with them, but at least the V4 Runner warned people about it so if they carried on and ran into errors, it was their own fault. With the V4 to V5 auto-conversion, there's no warning at all. Someone playing the game and running into an error isn't going to know whether this is because of the auto-conversion or because of a fault on behalf of the game's author, but either way they're going to be left with a very negative opinion of the game as a result. I'm guessing that most visitors to the site will choose to play the games via the WebRunner, which auto-converts with no warning, rather than downloading and installing ADRIFT, which means that they're going to run into problems.

And, perhaps the biggest issue for me right now, is that I don't want to have to go through double the amount of testing due to this. I don't want to write my game with V4, test it with V4, then have to test it again with V5. Testing isn't a whole lot of fun at the best of times, and doubling the amount of testing doesn't fill me with much enthusiasm. Aside from anything else, I'm not that familiar with V5 anyway so there's no guarantee I could fix any of the issues that arose.

[rotter]

This is a big problem, I had a really bad review of one of my game Unauthorised Termination. Turns out the V5 runner really doesn't like the game and parts of it simply do not work (teleport). I guess it could be sloppy programming on my part. But, the original game survived the IFComp process using the V4 runner. Like David I find this very worrying.

[Kennedy]

It seems the web runner no longer to load version 4 TAF files I've noticed?

[David Whyld]

It isn't even loading V5 games by the look of it.

[Lazzah]

It isn't even loading V5 games by the look of it.

I just tried to play a couple of games online and you are right - nothing works! I think Campbell is on holiday, he mentioned on Facebook he had booked a holiday in Alicante (I think). He told me he was having a break from ADRIFT. As he hasn't visited the forum since the early hours of last Thursday, I think we can safely assume nothing will be done about the problem for a while!

Edit: He's around, but just lurking!

[Campbell]

Which games don't load? The three linked on the WebRunner front page all work ok.

[Lazzah]

Campbell! Are you back?! Hope you had a nice break. I tried three different games yesterday, including my own "Fortress of Fear", and when you clicked on the green "Play Online" triangle, nothing happened, the page didn't even start to load at all. It seems to be working fine today!

[David Whyld]

Could we please have an answer to this, Campbell? It's not doing anyone - least of all you - any favours if people come away from ADRIFT with a negative impression because they're playing a game with the wrong version of the software. It's frustrating enough being criticised for making mistakes in your game, but even worse when the mistakes aren't even your fault.

[Kennedy]

Maybe put a new feature request on the bug tracker that when opening version 4 files in the runner or editor that a warning message is displayed?

[Campbell]

I'd still rather have a look at the problem(s) first, if you could send me that example.

[David Whyld]

Okay, what's your e-mail address?

[Campbell]

How long have you known me now?